Applied Mechanics and Materials Vols. 236-237

Abstract: China telecommunications market is becoming more competitive, the operators are facing the severe costumer churn problem, and how to predict and effectively reduce the costumer churn directly concerns the survival and development of every operator. Therefore, the least squares support vector machine(LS-SVM) algorithm is adopted to build customer churn model, mainly including data cleaning, normalization, building forecasting model, model prediction, etc. The case study shows that the customer churn prediction using the LS-LSV has high precision, small error and remarkable effect.

Abstract: In this paper, a novel approach, namely local variation projection (LVP), is presented for face recognition. LVP defines an adjacency graph to model the variation among nearby face images, which includes the within-class variation and between-class variation, also called margin. In order to better detect the discriminant structure, we assign a small weight to the variation among nearby face images from the same class. Based on this content, a concise feature extraction criterion is built for dimensionality reduction. Experiments indicate the effectiveness of our proposed approach.

Abstract: A specific form of the spectral-domain Green’s functions for the superstrate-loaded cylindrically conformal microstrip structure is developed through the rigorous full-wave analysis. The incident fields produced by the known radial current source along the thin probe are also deduced in this special structure for the first time. Ultimately, with these considerations above, the surface current distribution can be obtained by the Method of Moment.

Abstract: A segmented loop antenna is presented for ultra-high frequency (UHF) near-field radio frequency identification (RFID) reader. The segmented antennas have been constructed that can operate at resonant diameters while still providing good magnetic coupling. The antenna printed on a FR4 printed circuit board (PCB) with an overall size of 170× 170 × 1 mm3 achieves good impedance matching and uniform magnetic field distribution over an operating bandwidth of 840–1120 MHz, which is desirable for UHF near-field RFID reader applications.

Abstract: Due to the water liquidity and undulation in irrigation channel as well as the vary kinds of external disturbances, the value of water change couldn’t be smooth. The signal of water level measuring is a nonlinear, non-stationary signal. In order to improve the measuring precision of water level, in the processing of water level measurement signal, the algorithm of Empirical Mode Decomposition is introduced. It can optimize the water level measured values and reduce the influence of the water wave to the water level measurement. Experiment shows this method is effective to improve the measuring precision of water level.

Abstract: Propose an application of multi-objective genetic algorithm (GA) for pattern synthesis of circular arrays with equally spaced isotropic elements. Aim at reduce the side-lobe level and deep the null of array antenna patterns, adopt improved real-coded genetic algorithm, which modify the crossover operator and mutation operator compare with standard GA, both the crossover and mutation probability adjustable in accordance with the generation numbers, in the final, the approach for antenna pattern synthesis can suppress interferences by deep the nulls at directions of interfering sources and placing main beam in direction of desired signal by controlling amplitude weights. Numerical results show effectiveness of the procedure proposed.

Abstract: The pulse waveform design method for ultra wideband (UWB) communication based on particle swarm optimization (PSO) technique is discussed, and the performance of UWB system with designed pulse is analyzed. According to analysis, the design of pulse is transformed into a process of finding the minimal solution of a function about weighting coefficient. The weighting coefficients are made certain based on PSO algorithm. The properties such as bit error rate (BER) performance in both single link and multi-user, link budget of the UWB system employing the designed pulse are simulated. Compared with Scholtz’s monocycle and random combined pulse, the results showed that the pulse designed based on the PSO algorithm has better BER performance and communication performance.

Abstract: MIPS processor is widely used in embedded system, whose memory hierarchy has such an important effect on the performance and cost that how to design memory hierarchy rapidly is a key problem. To deal with the MIPS memory hierarchy research, we design a MIPS simulator “QtMips”. It is a GUI, Qt-based simulator for MIPS assembly language, which not only collects and analyzes real-time simulation information, but also supports fast memory configuration setting such as hierarchy setting, allocation setting, etc.

Abstract: Integrated temperature sensor of current mode is made based on the principle of relation between forward current of PN junction and temperature. It is of very good linearity, inter-exchange and accurate measurement, which has been brought to close attention of the industry recently. The paper focuses on the introduction of the principle of PTAT integrated temperature sensor of current mode and detailed analysis of the product of the type .In addition, the scheme, used for constant temperature control inside the furnace, is designed with the temperature control system consist of current mode integrated temperature sensor AD590.

Abstract: In order to improve an efficiency of energy detection for a spectrum sensing in cognitive radio (CR), this paper proposes a dynamic threshold optimization algorithm. The traditional energy detection algorithm uses a fixed threshold, and can't guarantee always the optimal sensing performance in any environment. The improvement for sensing performance need to minimize the undetected probability and the probability of false alarm, and it is dissimilar for different CR users to accept these two errors. We improve the traditional energy detection algorithm, and firstly introduce a preference factor to characterize CR users’ different requirements for these two errors, then, propose a dynamic threshold optimization algorithm by minimizing integrated detection error for different signal-to-noise ratio (SNR). The simulation results show that the proposed algorithm effectively reduces the integrated spectrum sensing error, and increases the probability of detection, especially in low SNR.